add cookie jar

Makefile

@@ -44,7 +44,7 @@ help:
 
 # $(ZIG) commands
 # ------------
-.PHONY: build build-dev run run-release shell test bench download-zig wpt unittest
+.PHONY: build build-dev run run-release shell test bench download-zig wpt unittest data
 
 zig_version = $(shell grep 'recommended_zig_version = "' "vendor/zig-js-runtime/build.zig" | cut -d'"' -f2)
 
@@ -199,6 +199,9 @@ install-zig-js-runtime:
 	@cd vendor/zig-js-runtime && \
 	make install
 
+data:
+	cd src/data && go run public_suffix_list_gen.go > public_suffix_list.zig
+
 .PHONY: _build_mimalloc
 
 MIMALLOC := $(BC)vendor/mimalloc/out/$(OS)-$(ARCH)

src/data/public_suffix_list_gen.go

@@ -0,0 +1,42 @@
+package main
+
+import (
+	"bufio"
+	"fmt"
+	"net/http"
+	"strings"
+)
+
+func main() {
+	resp, err := http.Get("https://publicsuffix.org/list/public_suffix_list.dat")
+	if err != nil {
+		panic(err)
+	}
+	defer resp.Body.Close()
+
+	var domains []string
+
+	scanner := bufio.NewScanner(resp.Body)
+	for scanner.Scan() {
+		line := strings.TrimSpace(scanner.Text())
+		if len(line) == 0 || strings.HasPrefix(line, "//") {
+			continue
+		}
+
+		domains = append(domains, line)
+	}
+
+	lookup :=
+		"const std = @import(\"std\");\n\n" +
+			"pub fn lookup(value: []const u8) bool {\n" +
+			"    return public_suffix_list.has(value);\n" +
+			"}\n"
+	fmt.Println(lookup)
+
+	fmt.Println("const public_suffix_list = std.StaticStringMap(void).initComptime([_]struct { []const u8, void }{")
+	for _, domain := range domains {
+		fmt.Printf(`    .{ "%s", {} },`, domain)
+		fmt.Println()
+	}
+	fmt.Println("});")
+}

src/storage/cookie.zig

@@ -4,10 +4,11 @@ const Allocator = std.mem.Allocator;
 const ArenaAllocator = std.heap.ArenaAllocator;
 
 const DateTime = @import("../datetime.zig").DateTime;
+const public_suffix_list = @import("../data/public_suffix_list.zig").lookup;
 
 pub const Jar = struct {
     allocator: Allocator,
-    cookies: std.ArrayListUnmanaged(u8),
+    cookies: std.ArrayListUnmanaged(Cookie),
 
     pub fn init(allocator: Allocator) Jar {
         return .{
@@ -23,47 +24,174 @@ pub const Jar = struct {
         self.cookies.deinit(self.allocator);
     }
 
+    pub fn add(
+        self: *Jar,
+        cookie: Cookie,
+        request_time: i64,
+    ) !void {
+        const is_expired = isCookieExpired(&cookie, request_time);
+        defer if (is_expired) {
+            cookie.deinit();
+        };
+
+        for (self.cookies.items, 0..) |*c, i| {
+            if (areCookiesEqual(&cookie, c)) {
+                c.deinit();
+                if (is_expired) {
+                    _ = self.cookies.swapRemove(i);
+                } else {
+                    self.cookies.items[i] = cookie;
+                }
+                return;
+            }
+        }
+
+        if (!is_expired) {
+            try self.cookies.append(self.allocator, cookie);
+        }
+    }
+
     pub fn forRequest(
-        self: *const Jar,
+        self: *Jar,
         allocator: Allocator,
-        request_start: i64,
+        request_time: i64,
         origin_uri: ?Uri,
         target_uri: Uri,
         navitation: bool,
     ) !CookieList {
-        const is_secure = std.mem.eql(u8, target_uri.scheme, "https");
+        const target_path = target_uri.path.percent_encoded;
         const target_host = (target_uri.host orelse return error.InvalidURI).percent_encoded;
+
         const same_site = try areSameSite(origin_uri, target_host);
+        const is_secure = std.mem.eql(u8, target_uri.scheme, "https");
+
+        var matching: std.ArrayListUnmanaged(*Cookie) = .{};
 
         var i: usize = 0;
         var cookies = self.cookies.items;
         while (i < cookies.len) {
             const cookie = &cookies[i];
-            if (cookie.isExpired(request_start)) {
-                self.swapRemove(i);
+
+            if (isCookieExpired(cookie, request_time)) {
+                cookie.deinit();
+                _ = self.cookies.swapRemove(i);
                 // don't increment i !
                 continue;
             }
             i += 1;
 
             if (is_secure == false and cookie.secure) {
+                // secure cookie can only be sent over HTTPs
                 continue;
             }
 
-            // www.google.com
-
-            if (navitation == false and cookie.same_site != .strict) {
-                continue;
+            if (same_site == false) {
+                // If we aren't on the "same site" (matching 2nd level domain
+                // taking into account public suffix list), then the cookie
+                // can only be sent if cookie.same_site == .none, or if
+                // we're navigating to (as opposed to, say, loading an image)
+                // and cookie.same_site == .lax
+                switch (cookie.same_site) {
+                    .strict => continue,
+                    .lax => if (navitation == false) continue,
+                    .none => {},
+                }
             }
+
+            {
+                const domain = cookie.domain;
+                if (domain[0] == '.') {
+                    // when explicitly set, the domain
+                    //  1 - always starts with a .
+                    //  2 - always is a suffix match (or examlpe)
+                    if (std.mem.eql(u8, target_host, domain[1..]) == false and std.mem.endsWith(u8, target_host, domain) == false) {
+                        continue;
+                    }
+                } else if (std.mem.eql(u8, target_host, domain) == false) {
+                    // when Domain=XYX isn't specified, it's an exact match only
+                    continue;
+                }
+            }
+
+            {
+                const path = cookie.path;
+                if (path[path.len - 1] == '/') {
+                    // If our cookie path is doc/
+                    // Then we can only match if the target path starts with doc/
+                    if (std.mem.startsWith(u8, target_path, path) == false) {
+                        continue;
+                    }
+                } else {
+                    // Our cookie path is something like /hello
+                    if (std.mem.startsWith(u8, target_path, path) == false) {
+                        // The target path has to either be /hello (it isn't)
+                        continue;
+                    } else if (target_path.len < path.len or (target_path.len > path.len and target_path[path.len] != '/')) {
+                        // Or it has to be something like /hello/* (it isn't)
+                        // it isn't!
+                        continue;
+                    }
+                }
+            }
+            // we have a match!
+            try matching.append(allocator, cookie);
         }
+
+        return .{ ._cookies = matching };
     }
 };
 
-// abc.lightpanda.io is the same site as lightpanda.io or 123.lightpanda.io
-// or spice.123.lightpanda.io
+pub const CookieList = struct {
+    _cookies: std.ArrayListUnmanaged(*Cookie),
+
+    pub fn deinit(self: *CookieList, allocator: Allocator) void {
+        self._cookies.deinit(allocator);
+    }
+
+    pub fn cookies(self: *const CookieList) []*Cookie {
+        return self._cookies.items;
+    }
+};
+
+fn isCookieExpired(cookie: *const Cookie, now: i64) bool {
+    const ce = cookie.expires orelse return false;
+    return ce <= now;
+}
+
+fn areCookiesEqual(a: *const Cookie, b: *const Cookie) bool {
+    if (std.mem.eql(u8, a.name, b.name) == false) {
+        return false;
+    }
+    if (std.mem.eql(u8, a.domain, b.domain) == false) {
+        return false;
+    }
+    if (std.mem.eql(u8, a.path, b.path) == false) {
+        return false;
+    }
+    return true;
+}
+
 fn areSameSite(origin_uri_: ?std.Uri, target_host: []const u8) !bool {
     const origin_uri = origin_uri_ orelse return true;
     const origin_host = (origin_uri.host orelse return error.InvalidURI).percent_encoded;
+
+    // common case
+    if (std.mem.eql(u8, target_host, origin_host)) {
+        return true;
+    }
+
+    return std.mem.eql(u8, findSecondLevelDomain(target_host), findSecondLevelDomain(origin_host));
+}
+
+fn findSecondLevelDomain(host: []const u8) []const u8 {
+    var i = std.mem.lastIndexOfScalar(u8, host, '.') orelse return host;
+    while (true) {
+        i = std.mem.lastIndexOfScalar(u8, host[0..i], '.') orelse return host;
+        const strip = i + 1;
+        if (public_suffix_list(host[strip..]) == false) {
+            return host[strip..];
+        }
+    }
 }
 
 pub const Cookie = struct {
@@ -275,6 +403,215 @@ fn trimRight(str: []const u8) []const u8 {
 }
 
 const testing = @import("../testing.zig");
+test "cookie: findSecondLevelDomain" {
+    const cases = [_]struct { []const u8, []const u8 }{
+        .{ "", "" },
+        .{ "com", "com" },
+        .{ "lightpanda.io", "lightpanda.io" },
+        .{ "lightpanda.io", "test.lightpanda.io" },
+        .{ "lightpanda.io", "first.test.lightpanda.io" },
+        .{ "www.gov.uk", "www.gov.uk" },
+        .{ "stats.gov.uk", "www.stats.gov.uk" },
+        .{ "api.gov.uk", "api.gov.uk" },
+        .{ "dev.api.gov.uk", "dev.api.gov.uk" },
+        .{ "dev.api.gov.uk", "1.dev.api.gov.uk" },
+    };
+    for (cases) |c| {
+        try testing.expectEqual(c.@"0", findSecondLevelDomain(c.@"1"));
+    }
+}
+
+test "Jar: add" {
+    const expectCookies = struct {
+        fn expect(expected: []const struct { []const u8, []const u8 }, jar: Jar) !void {
+            try testing.expectEqual(expected.len, jar.cookies.items.len);
+            LOOP: for (expected) |e| {
+                for (jar.cookies.items) |c| {
+                    if (std.mem.eql(u8, e.@"0", c.name) and std.mem.eql(u8, e.@"1", c.value)) {
+                        continue :LOOP;
+                    }
+                }
+                std.debug.print("Cookie ({s}={s}) not found", .{ e.@"0", e.@"1" });
+                return error.CookieNotFound;
+            }
+        }
+    }.expect;
+
+    const now = std.time.timestamp();
+
+    var jar = Jar.init(testing.allocator);
+    defer jar.deinit();
+    try expectCookies(&.{}, jar);
+
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "over=9000;Max-Age=0"), now);
+    try expectCookies(&.{}, jar);
+
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "over=9000"), now);
+    try expectCookies(&.{.{ "over", "9000" }}, jar);
+
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "over=9000!!"), now);
+    try expectCookies(&.{.{ "over", "9000!!" }}, jar);
+
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "spice=flow"), now);
+    try expectCookies(&.{ .{ "over", "9000!!" }, .{ "spice", "flow" } }, jar);
+
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "spice=flows;Path=/"), now);
+    try expectCookies(&.{ .{ "over", "9000!!" }, .{ "spice", "flows" } }, jar);
+
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "over=9001;Path=/other"), now);
+    try expectCookies(&.{ .{ "over", "9000!!" }, .{ "spice", "flows" }, .{ "over", "9001" } }, jar);
+
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "over=9002;Path=/;Domain=lightpanda.io"), now);
+    try expectCookies(&.{ .{ "over", "9000!!" }, .{ "spice", "flows" }, .{ "over", "9001" }, .{ "over", "9002" } }, jar);
+
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "over=x;Path=/other;Max-Age=-200"), now);
+    try expectCookies(&.{ .{ "over", "9000!!" }, .{ "spice", "flows" }, .{ "over", "9002" } }, jar);
+}
+
+test "Jar: forRequest" {
+    const expectCookies = struct {
+        fn expect(expected: []const []const u8, list: *CookieList) !void {
+            defer list.deinit(testing.allocator);
+            const acutal_cookies = list._cookies.items;
+
+            try testing.expectEqual(expected.len, acutal_cookies.len);
+            LOOP: for (expected) |e| {
+                for (acutal_cookies) |c| {
+                    if (std.mem.eql(u8, e, c.name)) {
+                        continue :LOOP;
+                    }
+                }
+                std.debug.print("Cookie '{s}' not found", .{e});
+                return error.CookieNotFound;
+            }
+        }
+    }.expect;
+
+    const now = std.time.timestamp();
+
+    var jar = Jar.init(testing.allocator);
+    defer jar.deinit();
+
+    const test_uri_2 = Uri.parse("http://test.lightpanda.io/") catch unreachable;
+
+    {
+        // test with no cookies
+        var matches = try jar.forRequest(testing.allocator, now, test_uri, test_uri, true);
+        try expectCookies(&.{}, &matches);
+    }
+
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "global1=1"), now);
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "global2=2;Max-Age=30"), now);
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "path1=3;Path=/about"), now);
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "path2=4;Path=/docs/"), now);
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "secure=5;Secure"), now);
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "sitenone=6;SameSite=None;Path=/x/"), now);
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "sitelax=7;SameSite=Lax;Path=/x/"), now);
+    try jar.add(try Cookie.parse(testing.allocator, test_uri, "sitestrict=8;SameSite=Strict;Path=/x/"), now);
+    try jar.add(try Cookie.parse(testing.allocator, test_uri_2, "domain1=9;domain=test.lightpanda.io"), now);
+
+    {
+        // nothing fancy here
+        var matches = try jar.forRequest(testing.allocator, now, test_uri, test_uri, true);
+        try expectCookies(&.{ "global1", "global2" }, &matches);
+    }
+
+    {
+        // matching path without trailing /
+        var matches = try jar.forRequest(testing.allocator, now, test_uri, try std.Uri.parse("http://lightpanda.io/about"), true);
+        try expectCookies(&.{ "global1", "global2", "path1" }, &matches);
+    }
+
+    {
+        // incomplete prefix path
+        var matches = try jar.forRequest(testing.allocator, now, test_uri, try std.Uri.parse("http://lightpanda.io/abou"), true);
+        try expectCookies(&.{ "global1", "global2" }, &matches);
+    }
+
+    {
+        // path doesn't match
+        var matches = try jar.forRequest(testing.allocator, now, test_uri, try std.Uri.parse("http://lightpanda.io/aboutus"), true);
+        try expectCookies(&.{ "global1", "global2" }, &matches);
+    }
+
+    {
+        // path doesn't match cookie directory
+        var matches = try jar.forRequest(testing.allocator, now, test_uri, try std.Uri.parse("http://lightpanda.io/docs"), true);
+        try expectCookies(&.{ "global1", "global2" }, &matches);
+    }
+
+    {
+        // exact directory match
+        var matches = try jar.forRequest(testing.allocator, now, test_uri, try std.Uri.parse("http://lightpanda.io/docs/"), true);
+        try expectCookies(&.{ "global1", "global2", "path2" }, &matches);
+    }
+
+    {
+        // sub directory match
+        var matches = try jar.forRequest(testing.allocator, now, test_uri, try std.Uri.parse("http://lightpanda.io/docs/more"), true);
+        try expectCookies(&.{ "global1", "global2", "path2" }, &matches);
+    }
+
+    {
+        // secure
+        var matches = try jar.forRequest(testing.allocator, now, test_uri, try std.Uri.parse("https://lightpanda.io/"), true);
+        try expectCookies(&.{ "global1", "global2", "secure" }, &matches);
+    }
+
+    {
+        // navigational cross domain
+        var matches = try jar.forRequest(testing.allocator, now, try std.Uri.parse("http://example.com/"), try std.Uri.parse("http://lightpanda.io/x/"), true);
+        try expectCookies(&.{ "global1", "global2", "sitenone", "sitelax" }, &matches);
+    }
+
+    {
+        // non-navigational cross domain
+        var matches = try jar.forRequest(testing.allocator, now, try std.Uri.parse("http://example.com/"), try std.Uri.parse("http://lightpanda.io/x/"), false);
+        try expectCookies(&.{"sitenone"}, &matches);
+    }
+
+    {
+        // non-navigational same origin
+        var matches = try jar.forRequest(
+            testing.allocator,
+            now,
+            try std.Uri.parse("http://lightpanda.io/"),
+            try std.Uri.parse("http://lightpanda.io/x/"),
+            false,
+        );
+        try expectCookies(&.{ "global1", "global2", "sitenone", "sitelax", "sitestrict" }, &matches);
+    }
+
+    {
+        // exact domain match
+        var matches = try jar.forRequest(testing.allocator, now, test_uri, try std.Uri.parse("http://test.lightpanda.io/"), true);
+        try expectCookies(&.{"domain1"}, &matches);
+    }
+
+    {
+        // domain suffix match
+        var matches = try jar.forRequest(testing.allocator, now, test_uri, try std.Uri.parse("http://1.test.lightpanda.io/"), true);
+        try expectCookies(&.{"domain1"}, &matches);
+    }
+
+    {
+        // non-matching domain
+        var matches = try jar.forRequest(testing.allocator, now, test_uri, try std.Uri.parse("http://other.lightpanda.io/"), true);
+        try expectCookies(&.{}, &matches);
+    }
+
+    {
+        // cookie has expired
+        const l = jar.cookies.items.len;
+        var matches = try jar.forRequest(testing.allocator, now + 100, test_uri, test_uri, true);
+        try expectCookies(&.{"global1"}, &matches);
+        try testing.expectEqual(l - 1, jar.cookies.items.len);
+    }
+
+    // If you add more cases after this point, note that the above test removes
+    // the 'global2' cookie
+}
+
 test "Cookie: parse key=value" {
     try expectError(error.Empty, null, "");
     try expectError(error.InvalidByteSequence, null, &.{ 'a', 30, '=', 'b' });
@@ -461,7 +798,7 @@ fn expectCookie(expected: ExpectedCookie, url: []const u8, set_cookie: []const u
 }
 
 fn expectAttribute(expected: anytype, url: ?[]const u8, set_cookie: []const u8) !void {
-    const uri = if (url) |u| try Uri.parse(u) else dummy_test_uri;
+    const uri = if (url) |u| try Uri.parse(u) else test_uri;
     var cookie = try Cookie.parse(testing.allocator, uri, set_cookie);
     defer cookie.deinit();
 
@@ -475,8 +812,8 @@ fn expectAttribute(expected: anytype, url: ?[]const u8, set_cookie: []const u8)
 }
 
 fn expectError(expected: anyerror, url: ?[]const u8, set_cookie: []const u8) !void {
-    const uri = if (url) |u| try Uri.parse(u) else dummy_test_uri;
+    const uri = if (url) |u| try Uri.parse(u) else test_uri;
     try testing.expectError(expected, Cookie.parse(testing.allocator, uri, set_cookie));
 }
 
-const dummy_test_uri = Uri.parse("http://lightpanda.io/") catch unreachable;
+const test_uri = Uri.parse("http://lightpanda.io/") catch unreachable;